Jais 3 vs OpenAI API

Jais 3

Jais 3 runs on the owner's hardware through OpenAI-compatible servers in vLLM and SGLang, with tool calls, three reasoning levels and a 512K context, from a 2.2 GB 0.9B model to a 758 GB 375B one. The weights download without sign-up under Apache-2.0, but every launch command trusts remote code, and no technical report or security contact is published.

OpenAI API

Official OpenAPI document and an llms.txt index. Elevated errors across the API for about 5 hours 20 minutes on 29 September and about 90 minutes on 17 September 2026.

Jais 3

1. Start vLLM or SGLang with --reasoning-parser k2_horizon, and add --tool-call-parser k2_horizon (vLLM also --enable-auto-tool-choice) for tools, or tool calls come back as raw text
2. Pass reasoning_effort (high, medium or low) and tool_call_format inside chat_template_kwargs. Other values return a template error
3. Send each earlier assistant turn back with reasoning_content as a string, even an empty one. A missing or null thinking field makes the template raise an error
4. Send your own system message, or set default_system_prompt to an empty string. Without one the template inserts an identity prompt naming IFM, Inception42 and Cerebras
5. Budget at least 32,768 output tokens at high effort, which the cards recommend for every request

OpenAI API

1. Build on the Responses API. Astra calls tools only there
2. Use gpt-6-luna for routing and extraction and gpt-6.1-sol for most work. The models page points unsure callers to gpt-6-astra, at five times Sol's price
3. Move off the GPT-4o and GPT-4.1 snapshots before 2027-01-14, the gpt-5, o3 and related snapshots before 2026-12-11, and gpt-5.1, gpt-5.3-codex and gpt-5.4-nano before 2027-04-01. Check shutdown_date on GET /v1/models
4. Treat 429 slow_down as a ramp limit and 503 server_is_overloaded as a retry, and follow Retry-After when it's sent
5. Prompts over 272K tokens cost double on input. Trim before you pay for it